A python library for the fast and scalable computation of biologically meaningful individual specific networks

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• Individual Specific Networks (ISNs) infer biological relationships from omics data, offering functional insights.
• Existing methods struggle with the computational demands of large biological datasets.

Purpose of the Study:

• Introduce ISN-tractor, a Python library for efficient ISN construction and analysis.
• Address the need for scalable solutions for ISNs on large-scale omics data.

Main Methods:

• Developed ISN-tractor, a data-agnostic, optimized Python library.
• Benchmarked ISN-tractor against existing methods (e.g., LionessR) for time and memory efficiency.
• Applied ISN-tractor to diverse omics data types (transcriptomics, proteomics, genotype arrays) and real-world datasets (TCGA, HapMap).

Main Results:

• ISN-tractor exhibits superior scalability and efficiency compared to existing methods.
• Demonstrated successful application to large datasets like TCGA and HapMap.
• Identified distinct gene interaction patterns across and within cancer types.
• Filtration Curves revealed topological distinctions linked to clinical outcomes.
• Effectively clustered populations based on genetic relationships using Principal Component Analysis.

Conclusions:

• ISN-tractor provides a highly efficient and scalable solution for building and analyzing Individual Specific Networks.
• The library enables the application of ISNs to large-scale biological datasets, facilitating deeper insights into biological complexity.
• ISN-tractor's versatility supports diverse omics data types and applications, from cancer research to population genetics.